eoarchean crustal evolution of the jack hills zircon...

Available online at www.sciencedirect.com

www.elsevier.com/locate/gca

ScienceDirect

Geochimica et Cosmochimica Acta 146 (2014) 27–42

Eoarchean crustal evolution of the Jack Hills zircon sourceand loss of Hadean crust

Elizabeth A. Bell ⇑, T. Mark Harrison, Issaku E. Kohl, Edward D. Young

Dept. of Earth, Planetary, and Space Sciences, UCLA, United States

Received 10 March 2014; accepted in revised form 18 September 2014; available online 30 September 2014

Abstract

Given the global dearth of Hadean (>4 Ga) rocks, 4.4–4.0 Ga detrital zircons from Jack Hills, Narryer Gneiss Complex(Yilgarn Craton, Western Australia) constitute our best archive of early terrestrial materials. Previous Lu–Hf investigations ofthese zircons suggested that felsic (low Lu/Hf) crust formation began by �4.4 to 4.5 Ga and continued for several hundredmillion years with evidence of the least radiogenic Hf component persisting until at least �4 Ga. However, evidence for theinvolvement of Hadean materials in later crustal evolution is sparse, and even in the detrital Jack Hills zircon population, themost unradiogenic, ancient isotopic signals have not been definitively identified in the younger (<3.9 Ga) rock and zirconrecord. Here we show Lu–Hf data from <4 Ga Jack Hills detrital zircons that document a significant and previously unknowntransition in Yilgarn Craton crustal evolution between 3.9 and 3.7 Ga. The zircon source region evolved largely by internalreworking through the period 4.0–3.8 Ga, and the most ancient and unradiogenic components of the crust are mostly missingfrom the record after �4 Ga. New juvenile additions to the crust at ca. 3.9–3.8 Ga are accompanied by the disappearance ofunradiogenic crust ca. 3.9–3.7 Ga. Additionally, this period is also characterized by a restricted range of d18O after 3.8 Ga anda shift in several zircon trace element characteristics ca. 3.9–3.6 Ga. The simultaneous loss of ancient crust accompanied byjuvenile crust addition can be explained by a mechanism similar to subduction, which effects both processes on modern Earth.The oxygen isotope and trace element information, although less sensitive to tectonic setting, also supports a transition inzircon formation environment in this period.� 2014 Elsevier Ltd. All rights reserved.

1. INTRODUCTION: EMPIRICAL CONSTRAINTS ON

HADEAN–ARCHEAN TRANSITIONS

The nature of crust and the tectonic processes operatingon early Earth are difficult to constrain due to the fragmen-tary Eoarchean and essentially absent Hadean rock record(cf. O’Neil et al., 2008). In the absence of a paradigm forearly geodynamics, numerous authors have attempted toextend plausible models to early Earth conditions withdiffering conclusions (e.g., Davies, 1992, 2006; van Hunenand van den Berg, 2008; Sizova et al., 2010; Gerya, 2013).

http://dx.doi.org/10.1016/j.gca.2014.09.028

0016-7037/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +1 310 206 2940.E-mail address: [email protected] (E.A. Bell).

Alternatively, various lines of isotopic and mineral evidencefrom cratons have been interpreted to indicate substantialchanges in crustal evolution ca. 3 Ga, possibly connectedwith the onset of plate tectonics (Shirey and Richardson,2011; Dhuime et al., 2012; Naeraa et al., 2012; Debailleet al., 2013). However, the search for evidence of older tec-tonic regimes is limited by the dearth of samples and poten-tially compounded by the efficacy of crustal recycling byplate tectonics (e.g., Scholl and von Huene, 2010) or someother mechanism.

Despite the paucity of Hadean rocks, detrital zirconsfrom a pebble metaconglomerate at the “discovery site”

in the Jack Hills, Western Australia do include Hadeanmaterials, and range 4.4–3.0 Ga in age (e.g., Compstonand Pidgeon, 1986; hereafter this population is referred to


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.gca.2014.09.028&domain=pdf

28 E.A. Bell et al. / Geochimica et Cosmochimica Acta 146 (2014) 27–42

as the “Jack Hills zircons”). Various aspects of the >4 GaJack Hills zircons’ geochemistry have been used to infertheir formation in low-temperature, hydrous, granite-likemelting conditions (e.g., Mojzsis et al., 2001; Peck et al.,2001; Watson and Harrison, 2005; Harrison et al., 2008).In particular, previous work on the Lu–Hf isotopic system-atics of Jack Hills zircons demonstrated a dominantlyunradiogenic Hadean population (Amelin et al., 1999;Harrison et al., 2005, 2008; Blichert-Toft and Albarede,2008; Kemp et al., 2010) with isolation of low-Lu/Hf(enriched) reservoirs as early as 4.5 Ga and persistence ofthat material in the crust for at least several hundred mil-lion years (Harrison et al., 2008; Kemp et al., 2010). Highlyunradiogenic materials within error of the solar system ini-tial Hf isotopic composition were noted by Harrison et al.(2005, 2008) and persist until at least 4 Ga. The large rangein initial eHf (initial 176Hf/177Hf normalized to an assumedchondritic uniform reservoir, or CHUR) may suggest con-tinuous extraction, perhaps to �4.0 to 3.9 Ga (Harrisonet al., 2005; Blichert-Toft and Albarede, 2008).

The dominant Jack Hills age group at �3.6 to 3.3 Ga isgeochemically distinct from the broader Hadean popula-tion in several important respects, suggesting that animportant transition(s) occurred between 4.0 and 3.6 Gain the Yilgarn Craton crust. Zircons younger than 3.6 Gahave considerably more radiogenic Hf as a whole, suggest-ing a diminished influence of ancient Hadean crust in thezircon source area before 3.6 Ga (Amelin et al., 1999; Bellet al., 2011). In addition, some post-Hadean juvenile inputto the crust is required to explain the most radiogenic<3.6 Ga zircons (Bell et al., 2011). The Jack Hills oxygenisotope record also changes during the Eoarchean:although concordant Hadean zircons commonly populatethe range d18OSMOW �3& to 8&, with a few reportedhigher values (Mojzsis et al., 2001; Peck et al., 2001;Cavosie et al., 2005; Trail et al., 2007b), the <3.8 Ga popu-lation appears more mantle-like (cf. Peck et al., 2001; Belland Harrison, 2013; Bell et al., 2011). This probably reflectsthe sourcing of these younger zircons from magmas whichincluded fewer aqueously altered materials (Bell andHarrison, 2013; Bell et al., 2011). Scattered highly positiveand negative d18O values in the Eoarchean may also pointto post-crystallization disturbance.

Trace element-based indicators are also useful for mon-itoring the changing petrogenesis of the Jack Hills zircons;application of the Ti-in-zircon thermometer to the Hadeanpopulation revealed average crystallization temperatures(Txlln) �680 �C – similar to temperatures seen in hydrousgranitic magmas, and notably lower than the majority ofzircons from mantle-derived igneous complexes or terres-trial impact magmas (Watson and Harrison, 2005; cf. Fuet al., 2008; Harrison, 2009; Harrison et al., 2007;Hellebrand et al., 2007; Wielicki et al., 2012; Carley et al.,2014). Zircons from 4.0 to 3.3 Ga have Txlln similar to theHadean distribution (Bell et al., 2011), with the curiousexception of the period �3.91 to 3.84 Ga, in which a largegroup of concordant zircons yield a remarkably low aver-age apparent Txlln of �600 �C, with values extending tosubsolidus temperatures as low as 525 �C (Bell andHarrison, 2013). Other geochemical peculiarities of zircons

in this time period led Bell and Harrison (2013) to interpretthis distinct group as resulting from solid-state recrystalliza-tion (Hoskin and Black, 2000), likely due to a thermal eventin the zircon source terrane(s). The application of morecomprehensive trace element analyses to other 4.0–3.6 Gazircons, along with electron imaging and prior Ti-thermom-etry and oxygen isotope measurements (Bell and Harrison,2013) in this time period, provides a clearer picture of thefundamental nature of these samples (i.e., metamorphicvs. magmatic) and their role in crustal evolution.

The Jack Hills zircon population is poorly representedoutside of the two prominent age groups at 3.6–3.3 Gaand 4.2–3.8 Ga, so the true nature of its crustal evolutionhas remained uncertain. We present 160 new Lu–Hf–Pbisotopic measurements (by the coupled Hf–Pb protocol ofWoodhead et al., 2004) on 148 Jack Hills detrital zirconsranging in age from 4.2 to 3.3 Ga, with the majority datedto between 4.0 and 3.6 Ga. We also include measurementson 26 zircons from several nearby ca. 2.6 Ga granitoidsnot previously studied for Lu–Hf, including a previouslyunsampled microgranite 150 m from the discovery site,which further constrain the NGC crust’s Lu–Hf systematicsafter detrital zircon deposition. Increased sampling in theEoarchean clarifies the nature of the Hf distribution in thisperiod and demonstrates an important transition in crustalevolution whose timing was not evident from previous sam-pling. This new dataset, combined with data from previousLu–Hf studies of detrital and granitoid zircons at JackHills, allows for the tracing of NGC crustal evolution fromthe early Hadean to 2.6 Ga. We also present 49 new traceelement measurements on <4 Ga Jack Hills zircons andcompare the record of change in the Lu–Hf system to theoxygen isotope, trace element, and Ti thermometry recordsto further constrain the nature of these events in theEoarchean Yilgarn Craton crust.

2. METHODS

Zircons were selected for analysis from the sample set ofBell and Harrison (2013). They were previously dated byion microprobe either by Bell and Harrison (2013) orHolden et al. (2009). All detrital zircons were collected frompebble metaconglomerate at the discovery site. Bell andHarrison (2013) analyzed most samples for Ti and d18Oand thirty-one Eoarchean samples for rare earth elements(REE) and other trace elements. Granitoid zircons weretaken from the interior and margin of a granite intrudingthe metasediments ca. 3 km southwest of the discovery site(the “Blob” granite of Spaggiari et al., 2007). In addition,some zircons were also analyzed from a micrograniteintruding banded iron formation ca. 150 km southwest ofthe discovery site. More detailed information on sample sitelocations is given in Electronic Annex EA-1.

2.2. Lu–Hf–Pb measurements

We used backscattered electron and cathodolumines-cence images of our zircons to target the placement of nom-inally 69 lm diameter laser ablation pits made using aPhoton Machines 193 nm ArF ATL laser with a 4 Hz rep-

E.A. Bell et al. / Geochimica et Cosmochimica Acta 146 (2014) 27–42 29

etition rate coupled to a Thermo-Finnigan Neptune MC-ICPMS. Analyses were made over several days in Apriland May of 2013. We used the coupled Hf–Pb analysistechnique (distinct from the laser-ablation split-streamtechnique described by Fisher et al., 2014a) of Woodheadet al. (2004) to switch between measuring a Yb–Lu–Hf massset (171Yb, 173Yb, 174Yb/174Hf, 175Lu, 176Yb/176Lu/176Hf,177Hf, 178Hf, 179Hf) for Lu–Hf systematics to a Pb massset (206, 207, 208) for estimating age, using the analysissequence described by Bell et al. (2011). Briefly, thisinvolves measuring for 11 s on the Yb–Lu–Hf mass setand for 5 s on the Pb mass set in each analytical cycle;the first 2 s of each set were disregarded to allow for magnetsettling. We repeated this sequence for 15 cycles or until thezircon was ablated through, with most analyses continuingfor 10 or 11 cycles. We corrected data for our samples usingthe standard zircons AS3 (Paces and Miller, 1993) and MudTank (Black and Gulson, 1978). We used NIST-610 glass asa secondary Pb isotope standard, given its Pb isotopichomogeneity and lack of known matrix effect that wouldcomplicate its use as a standard for zircon analyses. Datafrom all standard analyses are graphed in Fig. 1 and tabu-lated in Electronic Annex EA-2. Data from all unknownanalyses are tabulated in EA-3, and representative CLimages shown in EA-4.

Detrital zircon ages presented for this study are thosemeasured during ICP-MS analysis, in order to more faith-fully match Hf isotopic compositions to zircon age.

Fig. 1. Standard analyses over all sessions. Solid lines indicate accepted va(A) 178Hf/177Hf for all zircon standard and unknown analyses after mAnczkiewicz (2004). (B) 176Hf/177Hf after mass-fractionation correction astandard after mass-fractionation correction and peak-stripping. (D) 207Pcorrection was not applied to the Pb isotopes, but in-session correction fastandard analyses.

Although the vast majority of zircons display 207Pb/206Pbages from LA-ICPMS and ion microprobe that agreewithin a few percent (see Fig. 2A, C, D), 13 analyses withEoarchean ion microprobe ages display >5% disagreement(Fig. 3). These analyses display LA-ICPMS ages both>4 Ga and <3.6 Ga. An additional 18 zircons displayed>4 Ga LA-ICPMS ages despite having 4.0–3.8 Ga ionmicroprobe ages. Given the inferred disturbance to somezircons’ U–Pb systems during the Eoarchean (Bell andHarrison, 2013), this suggests that some ion microprobeages in this period may represent only small surficialdomains or may be artifacts from zircon disturbance. How-ever, the majority of zircons in this period do display goodagreement between ages. The reverse situation, of Eoarch-ean LA-ICPMS ages among zircons with ion microprobeages outside this period, is not common in this dataset.

Only domains within the zircon that record a homoge-neous Pb isotopic age for at least three concurrent Hf–Pbcycles were considered. Another concern in the correctassignment of age to Hf isotopic composition is the possi-bility of ancient Pb loss, which can cause a lowering ofthe 207Pb/206Pb age with relatively little loss of concor-dance. Use of the younger apparent age would cause anartificial lowering of the zircon’s calculated eHf. Accord-ingly, most zircons used in this study were >95% concor-dant according to ion microprobe U–Pb analyses, andmany were less than 2% discordant (along with 17 < 90%concordant – listed in Electronic Annex EA2 and shown

lues, dashed lines indicate the average over the indicated standards.ass-fractionation correction, with accepted value of Thirlwall andnd peak-stripping for 176Yb and 176Lu. (C) 176Lu/177Hf for the AS3b/206Pb for the AS3 and NIST 610 standards. Mass fractionation

ctors were assigned to all standard and unknown analyses based on

Fig. 2. Hf–Pb samples in several potential indicators of age/Hf composition mismatch. 2r external error bars are shown. (A) Ion microprobe(“SIMS”) vs. LA-ICPMS age estimates with the 1:1 agreement line shown. Most samples agree within 2%. (B) our samples in eHf vs. age space,grouped by ion microprobe-ICPMS age disagreement. (C) Our samples in eHf vs. age space, grouped by the percent discordance of the ionmicroprobe age. (D) Analyses with ion microprobe ages less than 2% discordant grouped by percent disagreement between ion microprobeand ICPMS ages. The dashed line reflects, for reference, the evolution of a primordial felsic reservoir (176Lu/177Hf = 0.0125 following themodern upper continental crust value of Chauvel et al., 2014) and the dotted line that of a primordial mafic reservoir (176Lu/177Hf = 0.02).

Fig. 3. Detrital zircons with >10% age disagreement between theion microprobe and LA-ICPMS results. The age, eHf distributionsfor these grains using both the ion microprobe and LA-ICPMS207Pb/206Pb ages are shown. Error bars have been omitted tofacilitate comparison. Arrows are drawn from the ion microprobeage/eHf to the ICPMS age/eHf.


in Fig. 2C). There is little qualitative difference in the eHf vs.age distribution of the <2% discordant zircons (Fig. 2D) vs.the full sample set (Fig. 2B, C). Because we do not have ameasure for concordance during the LA-ICPMS analysis,

these ion microprobe concordance measurements cannotbe linked to the zircon Hf composition definitively, butwe consider the good agreement of ages between the tech-niques to show little room for systematic error based on thisuncertainty for this particular dataset.

Common Pb contamination is another concern in thecorrect matching of age with Hf composition, since it canlead to artificially higher 207Pb/206Pb ages. Although oursetup was unable to monitor 204Pb to assess common Pbcontamination we did measure 208Pb. Common Pb mani-fests in such measurements as a mixing in of radiogenicPb with a high-208Pb/206Pb, high-207Pb/206Pb end member(e.g., Blichert-Toft and Albarede, 2008). Fig. 4 shows ourdata in 208Pb/206Pb vs. 207Pb/206Pb. The low 208Pb/206Pband lack of obvious mixing with a common Pb end memberamong the detrital zircons stands in contrast to two of thesampled igneous units (samples from the interior and themargin of the “Blob” granite of Spaggiari et al., 2007).The former granitoid results are similar to apparent com-mon Pb contamination seen in an earlier solution ICPMSstudy of the Jack Hills detrital zircons (Blichert-Toft andAlbarede, 2008). We conclude that common Pb contamina-tion is not significant among our studied detrital zircons.Zircons from the local ca. 2.6 Ga granitoids (Pidgeon andWilde, 1998) are forced to an age of 2.67 Ga for the consid-

Fig. 4. LA-ICPMS Pb isotope data in 208Pb/206Pb vs. 207Pb/206Pb.Several metaigneous units of known age ca. 2.7 Ga show zirconcompositions skewing towards higher 208Pb/206Pb and 207Pb/206Pb– expected for common Pb contamination (see, e.g., Blichert-Toftand Albarede, 2008, for a similar example). Detrital Jack Hillszircons do not appear to be affected. Because of the apparentcontamination and the artificially old ages it causes, we will use anage of 2.67 Ga when considering the Hf isotopic compositions ofour meta-igneous zircons.


eration of their Hf isotopic compositions to avoid the arti-ficially older ages that result from the apparently wide-spread common Pb contamination.

We have omitted from our figures all data points fromthis study with 2r error bars >4e, but these 12 analysesdo not qualitatively change the eHf-age distribution andare tabulated along with the graphed data in EA-3. Wehave time-corrected our Hf isotope ratios using the 176Ludecay constant of Soderlund et al. (2004) and the CHURparameters of Bouvier et al. (2008). All data from previouswork are evaluated using the same parameters, sometimesrequiring a recalculation of eHf from the original study.

2.3. Trace elements

We carried out analyses for Ti, P, REE, Hf, U, and Thon 4.0–3.3 Ga zircons using the CAMECA ims1270 ionmicroprobe at UCLA in three sessions during December2008, January 2011, and May 2012 (i.e., before Lu–Hf–Pbanalysis). Primary O- beam intensities of �15 nA and a spotdiameter of �30 lm were used. Secondary ions weredetected at low MRP (m/Dm �2000) and high energy offset(�100 eV). NIST610 standard glass was used for calibra-tion. All data are presented in Electronic Annex EA-5.

3. RESULTS

Zircons in the period 4.0–3.6 Ga differ from the prevail-ing Hadean and <3.6 Ga Jack Hills zircon populations inseveral geochemical variables relevant to petrogenesis andcrustal history. Although �70% of zircons within the main4.2–3.8 and 3.6–3.3 Ga populations have U–Pb systems<10% discordant, within the 3.8–3.6 Ga age minimum only�50% of zircons are <10% discordant (data from Bell andHarrison, 2013). Zircons in the period 3.9–3.6 Ga are alsomore likely to be richer in Hf and U than the prevailingJack Hills zircon population.

3.1. Lu–Hf–Pb

Fig. 5 shows our data in eHf vs. age space, along withprevious Jack Hills detrital zircon Hf measurements(Amelin et al., 1999; Harrison et al., 2005, 2008; Blichert-Toft and Albarede, 2008; Kemp et al., 2010; Bell et al.,2011). Fig. 5A includes all Hf isotopic data from previousstudies of the zircons. The superchondritic results foundby Harrison et al. (2005) and Blichert-Toft and Albarede(2008) by solution ICP-MS and laser ablation ICP-MSanalyses for Lu–Hf tied to ion microprobe U–Pb ages havenot been replicated by later studies (including this study),which have found almost exclusively negative eHf values.Fig. 5B filters the database of previous results to those mea-sured using a concurrent Lu–Hf and Pb measurement tech-nique, which minimizes the danger of incorrect agecorrection of the Hf isotopic data. This filtered dataset isused for the remainder of the figures and discussion andincludes the data of Bell et al. (2011), Kemp et al. (2010),and most of the data of Harrison et al. (2008).

Our 4.0–3.8 Ga samples define a distribution similar tothat of the majority of Hadean zircons in both range andtrajectory in eHf vs. age space, with all but one sample dis-playing negative eHf. Neither the most radiogenic (withinerror of a projected depleted mantle evolution line) northe most unradiogenic (within error of the solar system ini-tial 176Hf/177Hf ratio, marked as “Forbidden” on figures)portions of the Hadean population are sampled by JackHills zircons after �4 Ga. With similar Hf–Pb samplingof the Hadean and Eoarchean (181 4.0–3.6 Ga zircons, vs.144 >4.0 Ga zircons in the database), it seems that isotopicremnants of this unradiogenic portion of the Hadean crustare at least much less prominent in the later record. After3.7 Ga, the zircon population at Jack Hills becomes strik-ingly more radiogenic, losing the most unradiogenic por-tion of the >3.8 Ga record as well as requiring post-Hadean juvenile input.

3.2. Trace elements

Rare earth element (REE) patterns for <4 Ga zircons(Fig. 6) resemble many terrestrial magmatic zircons (e.g.,Hoskin and Schaltegger, 2003), and display a steep HREE/LREE slope, low overall LREE, and prominent Ce and Euanomalies. High U contents above 500 ppm are generallymore common in the period 3.9–3.6 Ga, and the trace ele-ment ratios Th/U and Yb/Gd also show temporal variations(see Fig. 7). In particular low Th/U ratios between 4.0 and3.8 Ga may reflect zircon disturbance (Bell and Harrison,2013). On Figs. 6 and 7, U contents and Th/U ratios areage-corrected for decay, denoted by the subscript t.

4. DISCUSSION

The shift towards more radiogenic compositions amongthe younger zircons is further illustrated in Fig. 8, whichshows results from this and previous studies along withall >3.8 Ga Jack Hills zircon eHf results by the Hf–Pbmethod projected to 3.5 Ga along an upper crust-type res-ervoir (176Lu/177Hf = 0.0125; Chauvel et al., 2014). The

Fig. 5. (A) Our data plotted in eHf vs. crystallization age space along with a database of detrital zircons measured in previous studies ofArchean metasediments in the Jack Hills (JH) and nearby Mt. Narryer (MN) localities. 2r external error bars are shown. (B) The same plot,with the database of previous Jack Hills results restricted to those analyzed by the coupled Hf–Pb method. DM evolution curve for both plotscalculated by linearly projecting the current DM eHf of +18 to zero at 4.56 Ga. Data for previous Jack Hills detrital zircons from Amelin et al.(1999), Bell et al. (2011), Blichert-Toft and Albarede (2008), Harrison et al. (2005, 2008), and Kemp et al. (2010); data for Mt. Narryer detritalzircons from Nebel-Jacobsen et al. (2010). Previous Jack Hills coupled Hf-Pb data from Bell et al. (2011), Harrison et al. (2008), and Kempet al. (2010).

A B

C

Fig. 6. Post-Hadean REE results from this and several earlier studies of Jack Hills zircons. Most display classical terrestrial magmatic zirconpatterns (low LREE/HREE, pronounced Ce and Eu anomalies), although elevated LREE among some analyses may point to a degree ofalteration. Analyses from this study that overlapped cracks or inclusions are excluded. (A) 4.0–3.8 Ga zircons; (B) 3.8–3.6 Ga zircons; (C)<3.6 Ga zircons. Data from other studies are taken from Bell and Harrison (2013), Cavosie et al. (2006), Crowley et al. (2005), and Peck et al.(2001).


younger zircon population is strikingly more radiogenicthan would result from remelting of the Hadean crustalone. Origins in relatively evolved felsic magmas (as usedfor this projection) for most zircons are likely, based onother geochemical characteristics. These include the lowTi-in-zircon crystallization temperatures in the vast major-

ity of studied Jack Hills zircons (e.g., Watson and Harrison,2005) and the granitoid-like inclusion assemblages (e.g.,Mojzsis et al., 2001; Hopkins et al., 2008, 2010). The goodfit of low-Lu/Hf, felsic-type reservoirs to many of the>3.7 Ga Jack Hills zircons is therefore unsurprising.Although the origin of some zircons could be explained

C D

E F

A B

Fig. 7. Temporal variability of trace element contents and ratios among Jack Hills zircons along with their relationships to the Hf isotopicdistribution. (A) eHf vs. age in the Jack Hills zircons, grouped by U content. (B) eHf vs. age in the Jack Hills zircons, grouped by Th/U. (C) Uvs. age from this and previous studies. (D) Th/U vs. age from this and previous studies. (E) Hf vs. age from this and previous studies. (F) Yb/Gd vs. age from this and previous studies. U and Th/U are age-corrected for radioactive decay since zircon crystallization. Yb/Gd isnormalized to the chondritic ratio. The legend for panels C–F is shown in panel D. References for data from other studies are in Fig. 6caption. Ages plotted in panels C–F were measured by ion microprobe (denoted “SIMS”).


by the remelting of an ancient mafic reservoir (e.g., Kempet al., 2010) mixed with younger material, the totality ofthe evidence suggests otherwise. Because melting of themantle yields basalt (modeled in figures with176Lu/177Hf � 0.022), modeling the evolution of the earlyJack Hills crust with only felsic reservoirs is unlikely to cap-ture the entirety of its history. Any felsic reservoirs weinvoke will have resulted from a more complicated earlierhistory involving mantle melting at some stage(s), but wedemonstrate that some Hf isotopic compositions do requirederivation from a low-Lu/Hf (i.e., felsic) reservoir.

The coincidence between the discontinuities in the JackHills Hf isotopic record, the truncation of oxygen isotopecompositions at 3.8 Ga found by Bell and Harrison(2013), and subtle changes in the trace element composi-tion of the zircons points to the Eoarchean and especiallythe interval 3.9–3.7 Ga as an important period of transi-tion in the evolution of early Yilgarn Craton crust.However, in determining the significance of the Hfisotopic distribution in this time interval, it is importantto assess the potential effects of ancient Pb loss on theHf isotopic record.

Fig. 8. Data from this and previous studies as labeled in Fig. 5.Red, closed circles are P3.7 Ga zircon Hf compositions projectedforward to 3.5 Ga along an upper crust-type reservoir to simulatethe expected eHf range for the younger crust if it were onlycomposed of felsic Hadean crust. The highly unradiogenic portionof the >3.7 Ga crust is not well-represented in the younger crust.(For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)


4.1. Potential effects of Pb loss

Although recent Pb loss leads to U–Pb discordance withlittle change in the 207Pb/206Pb age, a zircon that has under-gone ancient Pb loss may have a partially to fully reset207Pb/206Pb age with little resulting U–Pb discordance inthe present day, depending on the timing of loss. Incorrectage assignment to Hf isotope measurements, particularly inthe case of sampling mixed age domains, can lead to largedeviations in the calculated eHf (e.g., as modeled byHarrison et al., 2005). Along the same lines, artificial lower-ing of the 207Pb/206Pb age through Pb loss leads to artifi-cially lowered eHf determinations (e.g., Vervoort, 2011;Kemp et al., 2012; Guitreau and Blichert-Toft, 2014;Fisher et al., 2014a). It is therefore important to assessthe possibility that ancient Pb loss in some Jack Hills zir-cons has led to artificially unradiogenic eHf, and we presentseveral models of Pb loss in Hadean zircons in Fig. 9. JackHills zircons measured for Lu–Hf systematics in both thisand previous studies (Amelin et al., 1999; Harrison et al.,2005, 2008; Kemp et al., 2010; Bell et al., 2011) have aver-age 176Lu/177Hf �0.0008 with a standard deviation of0.0006. We thus model Pb loss with an envelope of 2 eHf-age trajectories corresponding to 176Lu/177Hf of 0.002 andzero. We also plot the evolution of a TTG-like (i.e., tona-lite–trondjemite–granodiorite) reservoir (176Lu/177Hf =0.005; Guitreau et al., 2012) for comparison.

Pb loss has been proposed as a mechanism to generatesome of the unradiogenic eHf signatures among Jack Hillszircons (Kemp et al., 2010; Guitreau and Blichert-Toft,2014), with Kemp et al. (2010) proposing that unradiogenicsignatures suggested as evidence for felsic Hadean crust(Harrison et al., 2005, 2008) were artifacts of Pb loss.Fig. 9A and B show modeled Pb loss from a 4.2 Ga zirconresulting from remelting of such a long-lived mafic reservoirand subjected to Pb loss at either 3.3 Ga (A; a known meta-morphic event affecting the NGC; Myers, 1988) or 3.7 Ga

(B; age of oldest NGC units; Fletcher et al., 1988; Kinnyet al., 1988). Fig. 9C shows that production of 4.3–4.0 Gazircons within error of the initial solar system 176Hf/177Hf(Harrison et al., 2008) could only occur in an ultralowLu/Hf reservoir or by Pb loss from the early products ofsuch a reservoir (modeled here from the oldest possiblehypothetical zircon at ca. 4.56 Ga).

Such Pb loss events could lower zircon age and eHf with-out incurring significant U–Pb discordance and could theo-retically produce many of the moderately unradiogeniczircons in the Hadean and Eoarchean Jack Hills record.However, the lowering of Hadean zircons’ apparent207Pb/206Pb age by several hundred million years, asrequired of the Pb loss model, would necessitate significantdegrees of Pb loss (e.g., 60% or more for the most unradi-ogenic Eoarchean zircons). Although Pb loss events havebeen proposed for Jack Hills zircons during the Eoarchean(Trail et al., 2007a; Abbott et al., 2012; Bell and Harrison,2013), such high degrees of Pb loss is likely to alter theinternal texture of the zircons, especially the fine oscillatoryzonation that characterizes most recognized igneous tex-tures among Jack Hills zircons (e.g., Cavosie et al., 2005;Trail et al., 2007b). Zircons in our sample set that Belland Harrison (2013) suggested have undergone significantPb loss tend to be more radiogenic; they are shown bythe high-U, low-Th/U group seen at ca. 3.9–3.8 Ga onFig. 7A and B. Although some magmatically zonedEoarchean zircons displaying several hundred millionyears’ worth of ancient Pb loss were noted by Iizukaet al. (2009), these appear to be an unusual case and alsowere reported to display sector zonation instead of theoscillatory zonation seen among these zircons (which ismore susceptible to blurring during alteration). SignificantPb loss may, however, be consistent with the patchy, alteredtextures seen in most studied 3.9–3.6 Ga zircons (Fig. 10).The latest clustering of highly unradiogenic, oscillatoryzoned zircons at ca. 3.9 Ga points to this time as the lastunambiguous magmatic event in this crustal source’s his-tory, with much of the 3.9–3.7 Ga record potentiallyalthough not certainly altered by Pb loss and artificiallylowered eHf.

4.2. Crustal reservoirs

Overall, the Jack Hills eHf distribution is best explainedby the mixing of several crustal reservoirs. We identifypotential crustal components in Fig. 11 and Table 1, someof which appear to be absent from the zircon record after4.0 and 3.7 Ga. The 3.7 Ga step is particularly dramaticand is observed just before the zircon population becomesmarkedly more radiogenic. Before this point, a heteroge-neous distribution of mostly negative eHf points to diversecrustal origins in the Hadean. Fig. 11 shows the projectedevolution of both mafic and felsic materials formed ca.4.56 Ga (i.e., the oldest possible age). The occurrence of zir-cons more unradiogenic than these evolution lines necessi-tates their origins in either lower-Lu/Hf materials or bythe artificial lowering of apparent age due to Pb loss.

The most unradiogenic compositions identified by previ-ous studies are within error of the solar system initial

Fig. 9. Modeled potential explanations of the unradiogenic Hf isotopic compositions of Reservoir A and B zircons by Pb loss (or long-livedevolution of TTG-like reservoirs). In each plot, labels on the modeled points are written with the percent Pb loss on the top and the percent U–Pb discordance (206Pb/238U vs. 207Pb/206Pb ages) for the grain today resulting from such a Pb loss event and assuming no subsequent Pb loss(a best-case scenario, given the modern-day Pb loss seen among most Jack Hills zircons). (A and B) Models for formation of Reservoir Bgrains by Pb loss from Reservoir C zircons originally crystallized at 4.2 Ga. (C) Models for formation of Reservoir A showing both TTGreservoir evolution and Pb loss trajectories.

Fig. 10. Jack Hills detrital zircons in eHf vs. age space, grouped bytype of internal zonation (as shown by CL imaging), incorporatingimaging data from several studies. Grains from this study, Bellet al. (2011), and Bell and Harrison (2013) are classified as either“altered,” “magmatic,” “altered magmatic” (original magmaticzoning still apparent despite alteration), ambiguous. “BCZ” graindisplays broad concentric zones that might be primary broadoscillatory zonation or blurred, originally thinner oscillatoryzonation. Grains from Kemp et al. (2010) are classified into theirset of 16 “best” grains (characterized by unaltered oscillatoryzonation, high degree of concordance, and magmatic Th/U) and allothers.


176Hf/177Hf, with concordant U–Pb ages between 4.35 and4.0 Ga (Harrison et al., 2005, 2008). They require the isola-tion of materials with Lu/Hf well below that for averagemodern upper continental crust by �4.4 to 4.5 Ga(Harrison et al., 2005, 2008). We refer to this material asReservoir A. A very early isolated TTG-like reservoir with176Lu/177Hf �0.005 (based on TTG measurements byGuitreau et al., 2012) can account for most of these grains.An important alternative is the artificial lowering of some“Reservoir A” zircons’ apparent ages by ancient Pb loss,in which case the latest apparent extent of Reservoir A atca. 4 Ga would not necessarily have geological significance.

The large group of zircons which are less radiogenicthan the oldest possible (i.e., 4.56 Ga) mafic reservoir alsodemonstrate that part of the Hadean record requires anextended felsic history even apart from Reservoir A(Fig. 11). Zircons in this group which are too radiogenicto have been produced from Reservoir A alone we referto as products of Reservoir B. Involvement of ancient oryounger upper crust-type reservoirs, remelting of ancientbasaltic crust to form more felsic reservoirs (at or beforeca. 4.2 Ga for the youngest and most unradiogenic Reser-voir B zircons), or significant mixing with Reservoir Acould all explain the origin of this material. Reservoir Bappears to evolve by internal reworking (and/or mixing

Fig. 11. The zircon record modeled by a mixture of hypotheticalbasaltic and felsic reservoirs (see Table 1). Reservoir A materialsare within error of the forbidden region (alternatively consistentwith an ancient very low-Lu/Hf reservoir, e.g., TTG). Reservoir Bmaterials require significant ancient felsic history. The Hf system-atics of Reservoirs C and D are explainable by a variety ofcompositions. Reservoirs A and B were lost to the zircon recordduring the Hadean or Eoarchean, whereas Reservoir C may beexpressed among the youngest Jack Hills zircons and Reservoir Dis post-Hadean (if extracted from our modeled DM, then ca. 3.9–3.8 Ga). “UCC”: upper continental crust-type reservoir with176Lu/177Hf = 0.0125 (Chauvel et al., 2014); basaltic reservoir176Lu/177Hf = 0.02; “TTG” = tonalite-trondjemite-granodioritetype reservoir with 176Lu/177Hf = 0.005 (Guitreau et al., 2012).


between Reservoir A and more radiogenic materials)between their formation and 3.7 Ga, its last manifestationin the Jack Hills record. Ancient Pb loss during NGC mag-matic and metamorphic events in the period 3.3–3.7 Ga(Fletcher et al., 1988) could produce many of the ReservoirB zircon signatures from originally more radiogenic zir-cons, but as discussed in Section 4.1 the youngest and mostunradiogenic grains would require very high degrees of Pbloss for mid- to early-Archean events (see Fig. 9A, B) andbased on zircon internal textures (Fig. 10) this processmay not dominate the zircon record until after 3.9 Ga.

The origins of the more radiogenic >3.8 Ga zircons andthe more unradiogenic <3.8 Ga zircons are significantly lessconstrained than Reservoirs A and B, and are consistentwith both remelting of ancient mafic or more recent felsiccrust (or mixtures between such materials). We label thisReservoir C. There is no discrete separation in Hf isotopic

Table 1A description of our posited crustal reservoirs in the Jack Hills detrital ziMafic reservoirs are modeled with 176Lu/177Hf = 0.02, and felsic reserv(Chauvel et al., 2014). The TTG value of 0.005 (Guitreau et al., 2012) is

Reservoir Formation time 176Lu/177Hf

A >4.4 Ga 60.005B Perhaps >4.4 Ga; perhaps

younger additions<0.02; may mix with A a

C Perhaps >4.4 Ga; perhapsyounger additions

Mafic? Felsic? A mixture

D 3.9–3.8 Ga Mafic? Felsic? A mixture

composition between Reservoirs B and C, with the distinc-tion between them being the theoretical consideration oftheir relationship with an ancient mafic reservoir. Finally,the more radiogenic <3.7 Ga crust represents mixingbetween Reservoir C and another reservoir extracted priorto 3.7 Ga (Reservoir D). Detrital zircons from Mt. Narryer,another location in the NGC, reveal the presence of juvenilecrust at �3.9 to 3.8 Ga (Nebel-Jacobsen et al., 2010). Giventhe coincidence between the juvenile nature of these zirconsand the first appearance of more juvenile materials amongyounger Jack Hills zircons, it is likely that they sample crustderived from the same event.

Most of these reservoirs are consistent with sourcesidentified in previous studies of Jack Hills zircons, withthe exception of the highly unradiogenic Reservoir A,which is evident in Harrison et al. (2005, 2008) but not seenin Kemp et al. (2010). This is almost certainly due to thesmall number of >4 Ga zircons (n = 44 out of 68 total) ana-lyzed by Kemp et al. (2010) relative to that of Harrisonet al. (2005, 2008; n = 220; although n = 60 Hadean zirconsfor coupled Hf–Pb measurements in Harrison et al., 2008)combined with the very small fraction of the Hadean zirconpopulation represented by Reservoir A (ca. 3%; 7% ofHadean Harrison et al., 2008 coupled Hf–Pb measure-ments). A sample of 44 Hadean zircons would be expectedto produce only 1 member of Reservoir A, a minor discrep-ancy (although the 3 expected if considering only theHarrison et al., 2008 results may be of greater concern).

Models of Eoarchean crustal evolution for the Jack Hillssource crust must account for the Hf isotopic discontinuityaccompanying the loss of Reservoir B and appearance ofReservoir D at ca. 3.8–3.7 Ga. The last appearance at ca.4.0 Ga of zircons from sources within error of the solar sys-tem initial 176Hf/177Hf (Harrison et al., 2008) may also besignificant, although the concomitant shift toward morepositive eHf is not in evidence and the low number of Res-ervoir A samples make the timing of its disappearanceuncertain. The fate of Reservoirs C and D is also an inter-esting question, requiring samples younger than the JackHills detrital zircons themselves.

A broad survey of detrital zircon Hf isotope composi-tions in modern Yilgarn Craton drainages (Griffin et al.,2004) identified several zircon populations consistent withthe internal reworking of 3.8 Ga felsic crust until �2.6 Ga(see Fig. 12), although they derive from other regions of

rcon record, including approximate formation and residence times.oirs as the upper continental crust value of 176Lu/177Hf = 0.0125used for Reservoir A.

Behavior Persiststo

Residencetime

Mixed with B or C? �4 Ga �0.5 Gand C? Mixed with A or C? 3.7 Ga 0.5–0.8 Ga

? <3.7 Ga: mixed with B?>3.7 Ga: mixed with D

63.3 Ga P0.9 Ga

? Mixed with C ? ?

Fig. 12. Narryer Gneiss Complex (NGC) meta-igneous (Kempet al., 2010; this study) and detrital zircons along with 1.5–3.8 Gadetrital zircons within 10% of U–Pb concordia from moderndrainages in the Yilgarn Craton (Griffin et al., 2004). Unlabeledsymbols are as on Fig. 5. Red arrows represent the eHf, ageevolution trajectories for 3.8 and 4.2 Ga UCC-like reservoirs (i.e.,176Lu/177Hf = 0.0125), which bound the <3.7 Ga Jack Hills distri-bution. There is little evidence for felsic Hadean crustal involve-ment in the sources of <3 Ga zircons. “DZ” = “detrital zircons.”(For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)


the Yilgarn Craton and their relationship with crust sam-pled by the detrital zircons is uncertain. The compositionof the detrital zircon source is uncertain after 3.3 Ga dueto the relatively few grains sampled for Lu–Hf, but zirconsfrom 3.7 to 2.6 Ga NGC granitoids largely overlap the176Hf/177Hf of the 3.7–3.3 Ga detrital population (Kempet al., 2010). Zircons from the ca. 2.6 Ga granitoids in theNGC range between �5 and �20e (Kemp et al., 2010; thisstudy), overlapping with the wider Yilgarn Craton distribu-tion, but demonstrating the persistence of some moreancient or more felsic crust within the NGC (Fig. 12) and

Fig. 13. Multicollector ion microprobe oxygen isotope analyses vs. age fHarrison et al., 2008; Bell et al., 2011; Bell and Harrison, 2013; Trail et alerror of the mantle range (defined here as 4.7–5.9& relative to SMOW), tAs pointed out by Bell and Harrison (2013), deviations from the mantle rPeck et al. (2001) because this study was carried out using single collectograins which were on average higher than the mantle range, in contradicfigure also omits several data from Trail et al. (2007b) at ca. 3.9 Ga an(3.65 Ga, �6), (3.75 Ga, �2), and two at ca. (3.6 Ga, �1).

potentially representing the evolution of the detrital zirconsource itself. The episodic loss of ancient crust in this ter-rane appears to show an increase in crustal residence timeswith decreasing age: Reservoir A forms by at least ca.4.4 Ga (Harrison et al., 2005, 2008) and resides in the crustfor up to 0.5 Ga; Hadean Reservoir B is lost at 3.9–3.7 Ga(with interpretations possibly complicated by Pb loss);potentially younger Hadean Reservoir C crust is expresseduntil at least 3.3 and perhaps 2.6 Ga. This trend may reflectincreasing crustal stability in a cooling Earth.

4.3. Magma types and alteration history of the Jack Hills

source

In Fig. 13A we present previous ion microprobe multi-collector oxygen isotope results for Jack Hills zircons(Cavosie et al., 2005; Trail et al., 2007a; Harrison et al.,2008; Bell et al., 2011; Bell and Harrison, 2013) vs. age,demonstrating the lack of concordant zircons both signifi-cantly above and below the mantle range after ca. 3.8 Ga(Bell and Harrison, 2013) with the exception of severalgrains at 3.6 Ga. Peck et al. (2001) present single collectorion microprobe oxygen isotope data, including 16<3.6 Ga zircons with average d18O higher than the mantlevalue. However, uncertainties arising from the less reliabletechnique, smaller dataset, and lack of reproducibility ofseveral high-d18O analyses on several of their Hadeangrains in a later study (Cavosie et al., 2005) lead us to notinclude this dataset in our analysis. Fig. 13B shows d18Ovs. eHf for all samples that have been analyzed for both sys-tems (this study; Harrison et al., 2008; Bell et al., 2011).High-d18O zircons are not well-represented in this dataset.The very low d18O seen among some 4.1–3.6 Ga grainsare mostly represented among materials at or more radio-genic than �8e, corresponding to Reservoir C among the>3.7 Ga sample set and Reservoirs C or D after 3.7 Ga.

or the Jack Hills zircons from several studies (Cavosie et al., 2005;., 2007a,b). Analyses are subdivided into three groups: those withinhose significantly above this range, and those significantly below it.ange become much rarer after 3.8 Ga. This figure omits the data ofr ion microprobe. Their results include 32 analyses on 16 <3.6 Gation to the more numerous multicollector results shown here. Thisd +10, along with 4 data reported by Bell and Harrison (2013) at


While none of these data were found by later imaging tohave been collected on cracks (Bell and Harrison, 2013),lowered d18O has been noted in disturbed zircon regionsin the Jack Hills (Trail et al., 2007a) and this may be furtherevidence of alteration of zircons in this interval. Indeed,between 3.8 and 3.6 Ga, low d18O is mostly associatedwith discordant zircon domains. The mantle-like d18Oamong most of the more unradiogenic zircons and alsoamong the <3.7 Ga population suggests less disturbance(heterogeneous or low-d18O) or less sedimentary input(high-d18O).

The ca. 3.8 Ga period is also interesting in light of pre-vious trace element studies. Bell and Harrison (2013) iden-tified a group with distinctly low Ti, P, and LREE, high Hf,and coupled high U and low discordance at 3.91–3.84 Ga(“Group II”) which they attributed to recrystallization ofolder zircon in a metamorphic event, and its coincidencein time with other changes in the zircon record is intriguing.These samples are generally more radiogenic than other zir-cons during this time period (which are generally below �8e) and cluster at ca. 3.9–3.8 Ga and �2 to �5 e (high-U,low-Th/U cluster on Fig. 7A, B). Overall, trace elementsshow a great deal of similarity among zircons throughoutthe Jack Hills detrital record, but the period 3.9–3.6 Gadoes display more common occurrences of high-U(>500 ppm) and -Hf (>12,500 ppm) chemistries. Higher Ucontents may contribute to the higher rate of discordanceamong 3.8–3.6 Ga zircons if the zircons did not undergorecrystallization. In addition to recrystallization, thesechemical effects may also be imparted by magmatic evolu-tion – for instance, zircon U and Hf concentrations andthe Yb/Gd ratio generally rise and the Th/U falls as gran-itoid magmas evolve through fractional crystallization (e.g.,Claiborne et al., 2010). Compared to other time periods,3.8–3.7 Ga zircons have a restricted range in Th/U andYb/Gd, lacking the higher Yb/Gd (>30) and lower Th/U(<0.4) common in the rest of the record. Although the sim-ilar, low average Txlln of �700 �C throughout much of therecord (Bell and Harrison, 2013) does suggest mainly gran-itoid sources, changes to zircon chemistry after 3.8 Gaprobably points to a shift in magmatic sources or style. Sev-eral trace element ratios (Th/U and Yb/Gd) may suggestthese represent zircon growth in more juvenile melts,although high-U and -Hf grains may suggest otherwise.

This potential change in provenance at ca. 3.8 Ga basedon trace element chemistry is corroborated by the similartiming of the last abundant high d18O occurrence and bythe first appearance of more radiogenic crust in the JackHills Hf record (along with the juvenile signatures in con-temporaneous Mt. Narryer detrital zircons; Nebel-Jacobsen et al., 2010). It is likely that these three signalsare causally linked. The more evolved magmatic signal at�3.63 Ga accompanied by a few zircons with high d18Oprobably points to more felsic sources involved in magmaproduction, including supracrustal materials. Very lowd18O and unusual trace element contents consistent withalteration in the period 3.9–3.6 Ga may point to anincreased incidence of zircon alteration in this time period,but most of these effects seem to be expressed among themore radiogenic Reservoir C.

4.4. Tectonic implications

Although isotopic evidence for Hadean crust has beenidentified in several other locations around the planet(e.g., O’Neil et al., 2008, 2013; Iizuka et al., 2009), the abun-dance of >3.8 Ga materials preserved in the Jack Hills con-glomerate and the extensive geochemical investigations ofthe zircons in this population have made it thus far a uniquewindow into early Earth. Our potential Eoarchean crustalloss event is identifiable because of the extensive Hf isotopicanalyses on Jack Hills zircons, given that Reservoirs A andB are a minority of the population. The loss of these unrad-iogenic materials from the zircon record may be key tounderstanding the (lack of) preservation of Hadean cruston the planet today – for instance, if the transitions seenin the Jack Hills detrital population during this time periodrepresent a global event which destroyed or partiallydestroyed the pre-existing crust.

The discontinuity in the zircon Hf record at �3.9 to3.7 Ga is characterized by both the loss of Hadean felsiccrust and the addition of juvenile crust. Based on the Mt.Narryer detrital zircons, this probably involved melting ofthe depleted mantle (Nebel-Jacobsen et al., 2010). TheManfred Complex in the extant NGC, which comprisesthe remnants of a ca. 3.7 Ga layered mafic intrusion(Fletcher et al., 1988), is another indication of juvenileinput to the NGC crust at this time. Materials less radio-genic than a hypothetical primordial mafic reservoir arepresent until 3.9–3.7 Ga and absent thereafter, requiringeither the loss or the significant dilution of this unradiogen-ic material in the region of active reworking. Given theintroduction of more juvenile material into the NGC atca. 3.9–3.8 Ga, the causes of this Hf isotopic discontinuityand the probable alteration of zircons during this time per-iod (Bell and Harrison, 2013) may be linked. The shifts inzircon geochemistry after 3.8 Ga probably reflect a chang-ing magmatic scenario, with the lack of high-d18O zirconsreflecting less metasedimentary input into magmas. Thelack of lower Th/U and higher Yb/Gd signatures in the per-iod 3.8–3.7 Ga probably largely reflects zircon generationduring reworking of this more juvenile material (althoughthis interpretation is complicated by the higher incidenceof high-U and -Hf zircons in this period).

The Eoarchean Hf isotopic record at Jack Hills suggestscrustal evolution characterized by the simultaneousdestruction of older felsic crust and formation of juvenilecrust, and requires a mechanism(s) that can accomplishboth. Although the nature of Eoarchean lithosphere is spec-ulative, two potential scenarios for the Jack Hills zirconsource are a subduction-like environment and meteoriticbombardment (Hopkins et al., 2010; Bell and Harrison,2013). On the modern Earth, coupled crustal loss and gainare the major feature of Hf isotopic evolution in accretion-ary orogens (Collins et al., 2011). Subduction simulta-neously recycles crust (e.g. Scholl and von Huene, 2010)while introducing juvenile melts into the crust. In the Phan-erozoic, subduction-related orogens are often expressed inthe zircon Lu/Hf record as an excursion toward more posi-tive eHf and the loss of highly unradiogenic materials(Collins et al., 2011), similar to the Eoarchean Jack Hills


record. At �4.0 Ga, the last appearance of Reservoir Amay signal a similar loss of crustal material, although thesmall number of samples representing Reservoir A rendersinterpretations about the timing of loss more uncertain.Subduction initiation is a poorly understood process, buteither induced or spontaneous nucleation of subductionzones is likely to involve heating of the upper plate(Stern, 2004), consistent with other indicators of possiblemetamorphism and zircon disturbance.

Other mechanisms for producing this pattern of crustalevolution as reflected in the Hf isotopic distribution are dif-ficult to find on present-day Earth. A large enough meteor-ite impact (or multiple impacts, given that this is the timeperiod of the hypothesized Late Heavy Bombardment;Tera et al., 1974), obliterating part of the crust is an alterna-tive explanation. Meteorite impacts may also homogenizethe nearby crust (Darling and Moser, 2012), which for asmall enough volume of Reservoir B might account for itsdisappearance if homogenized with a large proportion ofmore radiogenic material. Such an event would also havesignificant thermal effects on surrounding crust (Abramovand Mojzsis, 2009). Bell and Harrison (2013) interpretedthe geochemistry of some ca. 3.9–3.8 Ga zircons as indica-tive of recrystallization, which were tentatively linked tothe Late Heavy Bombardment. Both Abbott et al. (2012)and Trail et al. (2007a) found ca. 4.0–3.8 Ga rims onHadean zircons, also consistent with heating at this time.The possibly disturbed d18O and clearly disturbed internaltextures of many Eoarchean zircons are also consistent withmetamorphic alteration of zircons in this interval. However,the lack of a clear higher-Txlln signal among zircons in thistime period, a characteristic of impact melt-grown zircons(Wielicki et al., 2012), led Bell and Harrison (2013) to con-clude that possible Late Heavy Bombardment effects on thezircons were probably limited to far-field crustal heatingrather than a nearby impact that might be expected to oblit-erate or homogenize much of the Jack Hills source crust.Sufficiently energetic distant impacts might be sufficient toheat or potentially melt crust in the Jack Hills zircon source,but it is unclear how they would accomplish crustalrecycling or homogenization and juvenile melt addition atdistance. While exogenic sources of heating and zircon dis-turbance are consistent with some features of EoarcheanJack Hills zircons, it is not clear that they would accountfor the loss of Reservoir B. It is also unclear whether other,endogenic mechanisms of crust homogenization would leadto the apparent abrupt loss of Reservoir B rather than agradual decline in its prominence in the zircon record.

Of these two scenarios, we consider the most likelymechanism to create this Eoarchean Hf isotopic discontinu-ity to be a destructive plate boundary process operating atca. 3.9–3.7 Ga, destroying ancient crust while generatingjuvenile crust. The existence of subduction-like processesduring the Eoarchean – and even their viability in theNeoproterozoic – is contentious (see Stern, 2007). Thehigher heat content of the early Earth, due to higher radio-activity and accretional energy, would undoubtedly haveinfluenced mantle convection and its expression on the lith-osphere. Models variously support (e.g., Davies, 2006; vanHunen and van den Berg, 2008; Korenaga, 2013) or contra-

vene (e.g., Davies, 1992; Gerya, 2013) the role of early sub-duction on a warmer Earth. Some models suggest insteadthe existence of quasi-subduction regimes involving shallowunderthrusting of oceanic crust (Sizova et al., 2010) or onlyshort episodes of intermittent subduction (e.g., O’Neillet al., 2007; van Hunen and van den Berg, 2008) duringthe Archean. Empirical evidence has been limited due tothe sparse Archean and absent Hadean rock records,although thermobarometry of mineral inclusion assem-blages in Jack Hills zircon have been interpreted as evidencefor an underthrusting, subduction-like regime at 4.2–4.0 Ga(Hopkins et al., 2008, 2010). Hf isotopic evidence cannotdistinguish among these subduction-like scenarios, beyonddemonstrating crustal loss and gain. Evidence for distur-bance of some Eoarchean Jack Hills zircons suggests meta-morphic heating (mostly evident in the more radiogenicReservoir C), but whether this was connected to the eventssurrounding crustal loss and addition at 3.9–3.7 Ga isuncertain. The loss of Hadean crust in two apparent stepsat 4.0 and 3.7 Ga and the relatively short-lived period ofjuvenile input at ca. 3.8–3.7 Ga are suggestive of episodicrather than continuous crustal loss and juvenile additionin the NGC during the Eoarchean. However, since the JackHills zircons represent an unknown portion of the Archeancrust, our data cannot distinguish between episodic, short-lived local crustal overturn – if subduction-like, then similarto many convergent boundaries today – and the continuousoperation of such a mechanism during the Archean but indifferent regions around the planet.

5. CONCLUSIONS

The Lu–Hf systematics of Jack Hills zircons indicate animportant transition in crustal evolution during theEoarchean. Crust formed before 4 Ga evolved by internalrecycling and mixing among various reservoirs (potentiallyobscured by ancient Pb loss among some zircons) until 3.9–3.8 Ga, when the appearance of more radiogenic materials(mirrored at the Mt. Narryer site, Nebel-Jacobsen et al.,2010) indicates new juvenile addition to the crust. Muchof the Hadean crust was lost from the zircon recordbetween 3.9 and 3.7 Ga, and <3.7 Ga zircon eHf composi-tions are consistent with mixing between the remainingmore radiogenic Hadean crust and new juvenile addition.The coincident loss of ancient crust and input of juvenilecrust is plausibly but tentatively explained by a destructiveplate-boundary process, suggesting the operation of someform of plate tectonics at least by the Eoarchean. The lossof ancient crust at ca. 4 Ga may point to a similar episode,but the small number of samples with which this ancientreservoir is represented limits confidence in the timing ofits disappearance. Comparison of ancient Narryer GneissComplex zircons from detrital and igneous units with detri-tal zircons in the modern Yilgarn Craton reveals thatHadean crust was lost from the Yilgarn Craton in stepwisefashion, much of it within 0.5–0.8 Ga of Earth’s formation.If these crustal loss events were more widespread than theYilgarn Craton then they would have important implica-tions for the preservation of Hadean crust during Archeanand later tectonic events.


ACKNOWLEDGMENTS

Microgranite sample JHO3008 was provided by Steve Mojzsis.We thank Axel Schmitt and Rita Economos for invaluable instruc-tion and help in analyzing our samples on the ion microprobe.Reviews by Simon Wilde, Jeff Vervoort, and an anonymousreviewer greatly improved this manuscript. Comments by threeanonymous reviewers on an earlier version of this manuscript alsoimproved it greatly. The ion microprobe facility at UCLA is partlysupported by a grant from the Instrumentation and Facilities Pro-gram, Division of Earth Sciences, National Science Foundation.This research was conducted with support from a grant toT.M.H. from NSF-EAR’s Petrology/Geochemistry Program andan NSF Graduate Research Fellowship to E.A.B.

APPENDIX A. SUPPLEMENTARY DATA

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.gca.2014.09.028.

REFERENCES

Abbott S. S., Harrison T. M., Schmitt A. K. and Mojzsis S. J.(2012) Ti–U–Th–Pb depth profiling of Hadean zircons: a searchfor evidence of ancient extraterrestrial impacts. Proc. Natl.

Acad. Sci. USA 109, 13486–13492.Abramov O. and Mojzsis S. J. (2009) Microbial habitability of the

Hadean Earth during the late heavy bombardment. Nature 459,419–422.

Amelin Y. V., Lee D. C., Halliday A. N. and Pidgeon R. T. (1999)Nature of the Earth’s earliest crust from hafnium isotopes insingle detrital zircons. Nature 399, 252–254.

Bell E. A. and Harrison T. M. (2013) Post-Hadean transitions inJack Hills zircon provenance: a signal of the late heavybombardment? Earth Planet. Sci. Lett. 364, 1–11.

Bell E. A., Harrison T. M., McCulloch M. T. and Young E. D.(2011) Early Archean crustal evolution of the Jack Hills zirconsource terrane inferred from Lu–Hf, 207Pb/206Pb, and d18Osystematics of Jack Hills zircons. Geochim. Cosmochim. Acta

75, 4816–4829.Black L. P. and Gulson B. L. (1978) The age of the mud tank

carbonatite, strangways range, northern territory. BMR J.

Aust. Geol. Geophys. 3, 227–232.Blichert-Toft J. and Albarede F. (2008) Hafnium in Jack Hills

zircons and the formation of the Hadean crust. Earth Planet.

Sci. Lett. 265, 686–702.Bouvier A., Vervoort J. D. and Patchett P. J. (2008) The Lu–Hf and

Sm–Nd isotopic composition of CHUR: constraints fromunequilibrated chondrites and implications for the bulk compo-sition of terrestrial planets. Earth Planet. Sci. Lett. 273, 48–57.

Carley T. L., Miller C. F., Wooden J. L., Padilla A. J., Schmitt A.K., Economos R. C., Bindemann I. N. and Jordan B. T. (2014)Iceland is not a magmatic analog for the Hadean: evidencefrom the zircon record. Earth Planet. Sci. Lett. 405, 85–97.

Cavosie A. J., Valley J. W., Wilde S. A. and E.I.M.F. (2005)Magmatic d18O in 4400–3900 Ma detrital zircons: a record ofthe alteration and recycling of crust in the Early Archean. Earth

Planet. Sci. Lett. 235, 663–681.Cavosie A. J., Valley J. W., Wilde S. A. and E.I.M.F. (2006)

Correlated microanalysis of zircon: trace element, d18O, andU–Th–Pb isotopic constraints on the igneous origin of complex>3900 Ma detrital grains. Geochim. Cosmochim. Acta 70, 5601–5616.

Chauvel C., Garcon M., Bureau S., Besnault A., Jahn B. and DingZ. (2014) Constraints from loess on the Hf–Nd isotopiccomposition of the upper continental crust. Earth Planet. Sci.

Lett. 388, 48–58.Claiborne L. L., Miller C. F. and Wooden J. L. (2010) Trace

element composition of igneous zircon: a thermal and compo-sitional record of the accumulation and evolution of a largesilicic batholith, Spirit Mountain, Nevada. Contrib. Mineral.

Petrol. 160, 511–531.Collins W. J., Belousova E. A., Kemp A. I. S. and Murphy J. B.

(2011) Two contrasting Phanerozoic orogenic systems revealedby Hf isotopic data. Nat. Geosci. 4, 333–337.

Compston W. and Pidgeon R. T. (1986) Jack Hills, evidence ofmore very old detrital zircons in Western Australia. Nature 321,766–769.

Crowley J. L., Myers J. S., Sylvester P. J. and Cox R. A. (2005)Detrital zircon from the Jack Hills and Mount Narryer,Western Australia: evidence for diverse >4.0 Ga source rocks.J. Geol. 113, 239–263.

Darling J. R. and Moser D. E. (2012) Impact induced crustaldifferentiation: new insights from the Sudbury structure. LPSC

XLIII.Davies G. F. (1992) On the emergence of plate tectonics. Geology

20, 963–966.Davies G. F. (2006) Gravitational depletion of the early Earth’s

upper mantle and the viability of early plate tectonics. Earth

Planet Sci. Lett. 243, 376–382.Debaille V., O’Neill C., Brandon A. D., Haenecour P., Yin Q.-Z.,

Mattielli N. and Treiman A. H. (2013) Stagnant-lid tectonics inearly Earth revealed by 142Nd variations in late Archean rocks.Earth Planet. Sci. Lett. 373, 83–92.

Dhuime B., Hawkesworth C. J., Cawood P. A. and Storey C. D.(2012) A change in the geodynamics of continental growth 3billion years ago. Science 335, 1334–1336.

Fisher C. M., Vervoort J. D. and DuFrane S. A. (2014a) AccurateHf isotope determinations of complex zircons using the “laserablation split stream” method. Geochem. Geophys. Geosyst. 15.http://dx.doi.org/10.1002/2013GC004962.

Fletcher I. R., Rosman K. J. R. and Libby W. G. (1988) Sm–Nd,Pb–Pb and Rb–Sr geochronology of the Manfred complex,Mount Narryer, Western Australia. Precambr. Res. 38, 343–354.

Fu B., Page F. Z., Cavosie A. J., Fournelle J., Kita N. T., Lackey J.S., Wilde S. A. and Valley J. W. (2008) Ti-in-zircon thermom-etry: applications and limitations. Contrib. Mineral. Petrol. 156,197–215.

Gerya T. V. (2013) Precambrian geodynamics: concepts andmodels. Gondwana Res., http://dx.doi.org/10.1016/j.gr.2012.11.008.

Griffin W. L., Belousova E. A., Shee S. R., Pearson N. J. andO’Reilly S. Y. (2004) Archean crustal evolution in the northernYilgarn Craton: U–Pb and Hf-isotope evidence from detritalzircons. Precambr. Res. 131, 231–282.

Guitreau M. and Blichert-Toft A. (2014) Implications of discor-dant U–Pb ages on Hf isotope studies of detrital zircons. Chem.

Geol. 385, 17–25.Guitreau M., Blichert-Toft J., Martin H., Mojzsis S. J. and

Albarede F. (2012) Hafnium isotope evidence from Archeangranitic rocks for deep-mantle origin of continental crust. Earth

Planet. Sci. Lett. 337–338, 211–223.Harrison T. M. (2009) The Hadean crust: evidence from >4 Ga

zircons. Annu. Rev. Earth Planet. Sci. 37, 479–505.Harrison T. M., Blichert-Toft J., Muller W., Albarede F., Holden

P. and Mojzsis S. J. (2005) Heterogeneous Hadean hafnium:evidence of continental crust by 4.4–4.5 Ga. Science 310, 1947–1950.



http://refhub.elsevier.com/S0016-7037(14)00582-1/h0005

















































































http://dx.doi.org/10.1002/2013GC004962









http://dx.doi.org/10.1016/j.gr.2012.11.008

http://dx.doi.org/10.1016/j.gr.2012.11.008





















Harrison T. M., Watson B. E. and Aikman A. B. (2007)Temperature spectra of zircon crystallization in plutonic rocks.Geology 35, 635–638.

Harrison T. M., Schmitt A. K., McCulloch M. T. and Lovera O.M. (2008) Early (P4.5 Ga) formation of terrestrial crust:Lu–Hf, d18O, and Ti thermometry results for Hadean zircons.Earth Planet. Sci. Lett. 268, 476–486.

Hellebrand E., Moller A., Whitehouse M. and Cannat M. (2007)Formation of oceanic zircons. Geochim. Cosmochim. Acta

Suppl. 71, A391.Holden P., Lanc P., Ireland T. R., Harrison T. M., Foster J. J. and

Bruce Z. (2009) Mass-spectrometric mining of Hadean zirconsby automated SHRIMP multi-collector and single-collectorU/Pb zircon age dating: The first 100,000 grains. Int. J. Mass

Spectrom. 286, 53–63.Hopkins M., Harrison T. M. and Manning C. E. (2008) Low heat

flow inferred from >4 Gyr zircons suggests Hadean plateboundary interactions. Nature 456, 493–496.

Hopkins M., Harrison T. M. and Manning C. E. (2010)Constraints on Hadean geodynamics from mineral inclusionsin >4 Ga zircons.. Earth Planet. Sci. Lett. 298, 367–376.

Hoskin P. W. O. and Black L. P. (2000) Metamorphic zirconformation by solid-state recrystallization of protolith igneouszircon. J. Metamorph. Geol. 18, 423–439.

Hoskin P. W. O. and Schaltegger U. (2003) The composition ofzircon and igneous and metamorphic petrogenesis. In (eds. J.M. Hanchar, P. W. O. Hoskin) Zircon. Rev. Mineral. 53. pp.343–385.

Iizuka T., Komiya T., Johnson S. P., Kon Y., Maruyama S. andHirata T. (2009) Reworking of Hadean crust in the Acastagneisses, northwestern Canada: evidence from in-situ Lu–Hfisotope analysis of zircon. Chem. Geol. 259, 230–239.

Kemp A. I. S., Wilde S. A., Hawkesworth C. J., Coath C. D.,Nemchin A., Pidgeon R. T., Vervoort J. D. and DuFrane S. A.(2010) Hadean crustal evolution revisited: new constraints fromPb–Hf isotope systematics of the Jack Hills zircons. Earth

Planet. Sci. Lett. 296, 45–56.Kemp A., Vervoort J. and Whitehouse M. (2012) Detrital zircon: a

foggy window into early Earth evolution. V.M. Goldschmidt

Conference, Montreal, Quebec, Canada, June 2012.Kinny P. D., Williams I. S., Froude D. O., Ireland T. R. and

Compston W. (1988) Early archaean zircon ages from orthog-neisses and anorthosites at Mount Narryer, Western Australia.Precambrian Res. 38, 325–341.

Korenaga J. (2013) Initiation and evolution of plate tectonics onEarth: theories and observations. Annu. Rev. Earth Planet. Sci.

41, 117–151.Mojzsis S. J., Harrison T. M. and Pidgeon R. T. (2001)

Oxygen-isotope evidence from ancient zircons for liquidwater at the Earth’s surface 4300 Myr ago. Nature 409, 178–181.

Myers J. S. (1988) Early archaean Narryer Gneiss Complex,Yilgarn Craton, Western Australia. Precambr. Res. 38, 297–307.

Naeraa T., Schersten A., Rosing M. T., Kemp A. I. S., Hofmann J.E., Kokfelt T. F. and Whitehouse M. J. (2012) Hafnium isotopeevidence for a transition in the dynamics of continental growth3.2 Gyr ago. Nature 485, 627–630.

Nebel-Jacobsen Y., Munker C., Nebel O., Gerdes A., Mezger K.and Nelson D. R. (2010) Reworking of Earth’s first crust:constraints from Hf isotopes in Archean zircons from Mt.Narryer, Australia. Precambr. Res. 182, 175–186.

O’Neill C., Lenardic A., Moresi L., Torsvik T. H. and Lee C.-T. A.(2007) Episodic Precambrian subduction. Earth Planet. Sci.

Lett. 262, 552–562.

O’Neil J., Carlson R. W., Francis D. and Stevenson R. K. (2008)Nd-142 evidence for Hadean mafic crust. Nature 321, 1828–1831.

O’Neil J., Boyet M., Carlson R. W. and Paquette J.-L. (2013) Halfa billion years of reworking of Hadean mafic crust to producethe Nuvvuagittuq Eoarchean felsic crust. Earth Planet. Sci.

Lett. 379, 13–25.Paces J. B. and Miller J. D. (1993) Precise U–Pb ages of Duluth

complex and related mafic intrusions, northeastern Minnesota:geochronological insights into physical, petrogenetic, andpaleomagnetic and tectonomagnetic processes associated withthe 1.1 Ga mid-continent rift system. J. Geophys. Res. 98,13997–14013.

Peck W. H., Valley J. W., Wilde S. A. and Graham C. M. (2001)Oxygen isotope ratios and rare earth elements in 3.3–4.4 Gazircons: ion microprobe evidence for high d18O continentalcrust and oceans in the Early Archean. Geochim. Cosmochim.

Acta 65, 4215–4299.Pidgeon R. T. and Wilde S. A. (1998) The interpretation of

complex zircon U-Pb systems in Archaean granitoids andgneisses from the Jack Hills, Narryer Gneiss Terrane, WesternAustralia. Precam. Res. 91, 309–332.

Scholl D. W. and von Huene R. (2010) Subduction zone recyclingprocesses and the rock record of crustal suture zones. Can. J.

Earth Sci. 47, 633–654.Shirey S. B. and Richardson S. H. (2011) Start of the Wilson Cycle

at 3 Ga shown by diamonds from subcontinental mantle.Science 333, 434–436.

Sizova E., Gerya T., Brown M. and Perchuk L. L. (2010)Subduction styles in the Precambrian: insight from numericalexperiments. Lithos 116, 209–229.

Soderlund U., Patchett J. P., Vervoort J. D. and Isachsen C. E.(2004) The 176Lu decay constant determined by Lu–Hf and U–Pb isotope systematics of Precambrian mafic intrusions. Earth

Planet. Sci. Lett. 219, 311–324.Spaggiari C. V., Pidgeon R. T. and Wilde S. A. (2007) The Jack

Hills greenstone belt, Western Australia Part 2: lithologicalrelationships and implications for the deposition of �4.0 Gadetrital zircons. Precambr. Res. 155, 261–286.

Stern R. J. (2004) Evidence from ophiolites, blueschists, andultrahigh-pressure metamorphic terranes that the modernepisode of subduction tectonics began in Neoproterozoic time.Geology 33, 557–560.

Stern R. J. (2007) When and how did plate tectonics begin?Theoretical and empirical considerations. Chin. Sci. Bull. 52,578–591.

Tera F., Papanastassiou D. A. and Wasserburg G. (1974) Isotopicevidence for a terminal lunar cataclysm. Earth Planet. Sci. Lett.

22, 1–21.Thirlwall M. F. and Anczkiewicz R. (2004) Multidynamic isotope

ratio analysis using MC-ICP-MS and the causes of secular driftin Hf, Nd and Pb isotope ratios. Int. J. Mass Spec. 235, 59–81.

Trail D., Mojzsis S. J. and Harrison T. M. (2007a) Thermal eventsdocumented in Hadean zircons by ion microprobe depthprofiles. Geochim. Cosmochim. Acta 71, 4044–4065.

Trail D., Mojzsis S. J., Harrison T. M., Schmitt A. K., Watson E.B. and Young E. D. (2007b) Constraints on Hadean zirconprotoliths from oxygen isotopes, REEs and Ti-thermometry.Geochem. Geophys. Geosyst. 8, Q06014.

van Hunen J. and van den Berg A. P. (2008) Plate tectonics on theearly Earth: limitations imposed by strength and buoyancy ofsubducted lithosphere. Lithos 103, 217–234.

Vervoort J. (2011) The caveats in the use of Hf model ages inprovenance (and other) studies. Geological Society of America

Abstracts with Programs, vol. 43, Paper no. 195841.







































































































































Watson E. B. and Harrison T. M. (2005) Zircon thermometerreveals minimum melting conditions on earliest Earth. Science

308, 841–844.Wielicki M. T., Harrison T. M. and Schmitt A. K. (2012)

Geochemical signatures and magmatic stability of impactproduced zircons. Earth Planet. Sci. Lett. 321–322, 20–31.

Woodhead J. D., Hergt J. M., Shelley M., Eggins S. and Kemp R.(2004) Zircon Hf isotope analysis with an excimer laser, depthprofiling, ablation of complex geometries, and concomitant ageestimation. Chem. Geol. 209, 121–134.

Associate editor: Alexander Nemchin











eoarchean crustal evolution of the jack hills zircon...

Documents